LlamaIndex (Python)

The data framework for LLM apps. LlamaIndex turns unstructured data into agent-queryable knowledge — document loaders, vector indices, query engines, tools, and multi-agent orchestration via llama-agents.

Install

pip install llama-index llama-index-core

Version

llama-index-core v0.14.21 · April 2026 · Python 3.10+

Best for

RAG systems, document Q&A, enterprise knowledge agents.

Where to start

Need a RAG system in 10 minutes? Start at Creating basic ReAct agents + Query engine tools. Doing production-scale ingestion? Jump to Advanced implementations.

Zero → Hero

1 · Installation & setup Install, configure LLM + embedding · 15 min

2 · Architecture overview Stack layers, core abstractions · 15 min

3 · Core concepts Documents, nodes, indices, retrievers · 25 min

4 · Agent classes & types ReActAgent, FunctionCallingAgent · 20 min

5 · LLM & embedding config Providers, customisation · 15 min

6 · Basic ReAct agents First agent walkthrough · 20 min

7 · Tool integration FunctionTool, QueryEngineTool · 20 min

8 · Query engine tools RAG as a tool · 20 min

9 · Streaming responses Token-level streaming · 10 min

10 · llama-agents package Multi-agent package · 15 min

11 · Multi-agent orchestration Orchestrator, task queues · 25 min

12 · Advanced implementations Production patterns · 30 min

13 · Production Deployment, scaling, cost · 30 min

Jump to a topic

Core concepts

Agent classes

ReAct agents

FunctionTool

QueryEngineTool

LlamaHub tools

Multi-agent orchestration

Distributed systems

Advanced agents

Observability

Streaming server

Reference

Comprehensive guide Single-page technical reference.

Advanced agents

Advanced implementations

Observability

Recipes

Diagrams

Production guide

Streaming server (FastAPI)

What’s new in v0.14.20 (April 2026)

• Stabilised llama-agents package for multi-agent orchestration.
• Enhanced streaming support with token-level events.
• Expanded LlamaHub integrations (100+ data connectors).
• Improved query-engine-as-tool patterns.

Install, minimal example

from llama_index.core import VectorStoreIndex, SimpleDirectoryReader
from llama_index.core.agent import ReActAgent
from llama_index.core.tools import QueryEngineTool

documents = SimpleDirectoryReader("data").load_data()
index = VectorStoreIndex.from_documents(documents)

tool = QueryEngineTool.from_defaults(
    query_engine=index.as_query_engine(),
    name="knowledge_base",
    description="Answer questions about the documents",
)

agent = ReActAgent.from_tools([tool])
response = agent.chat("What are the key findings?")
print(response)

Ready? Start Zero → Hero Step 1 →

---

Revision history

Date	Version	Changes
2026-04-21	0.14.20	Index redesigned with Zero → Hero + Jump-to-topic grid.
April 3, 2026	0.14.20	llama-agents stabilisation; streaming; LlamaHub growth.
November 2025	0.14.8	Initial documented version.

Note: TypeScript implementation lives at LlamaIndex (TypeScript).